Ticket #562 (closed: fixed)
Muon: Asymetry vs field plots
| Reported by: | Nick Draper | Owned by: | Roman Tolchenov |
|---|---|---|---|
| Priority: | major | Milestone: | Iteration 17 |
| Component: | Keywords: | ||
| Cc: | Blocked By: | ||
| Blocking: | Tester: |
Description (last modified by Nick Draper) (diff)
We need to plan and discuss this one once you have had a look at the fortran files.
Test data for MUSR has been added to the C:\Mantid\Test\Nexus directory
Attached are two versions of the ‘asym’ program, ‘asym.for’ for single period data and ‘asym_rgxx_nexus.for’ . The former calculates the integral asymmetry for single period data, while the latter generates an integral asymmetry calculating the difference between two user selectable periods.
Muon data collection typically runs as a single period, where counts are accumulated into a set of histograms (one for each detector) until the run is stopped. There are, however, a growing number of experiments where the user wants to compare data recorded with and without an external stimulus (e.g. RF on/RF off). In this case, data collection is run as a number of periods, switching the appropriate external stimulus on/off and collecting data to a unique set of histograms for each period, and switching between periods after a fixed number of ISIS frames have been accumulated (typically 500). For a single stimulus two periods are defined, and data collection cycles around them many times during a typical run. For more complex experiments there may be two stimuli (e.g. RF and a laser) leading to four collection periods to explore all possible combinations. And so on. Typically an experiment would involve generating a set of runs with a variable such magnetic field scanned.
The attached programs provide two difference algorithms for calculating an integral asymmetry across a time window (actually defined in terms of histogram bins). INTSYM adds up all the forward and backward counts in the window (for both periods if required) and then calculates an asymmetry, while DIFSYM calculates an asymmetry at each time bin, combining these to arrive at some kind of average asymmetry. We’d like to preserve both methods.
The following is a list of features it would be good to have in the program:
- Combine functionality of ‘asym’ and ‘asym_rgxx_nexus’ into a single program.
- Run set read into a data buffer, allowing multiple window choices to be evaluated.
- The ability to add additional runs to the data buffer without rereading the full set would be great. (not available in the present program)
- Ideally offer a choice of scanned variable from the Nexus file. (not available in the present program)
- Handles variable number of histograms – 600 per period may soon be common. (restricted in the Fortran code)
- For multi-period data, a choice of which two periods are used to calculate asymmetries.
- Curve fitting (Lorentzian) for one or more peaks, with peak centres and linewidths fitted. (not available in the present program)
- The ability to define forward and backward histograms in a separate grouping table. (not available in the present program)
However, as a first step it would be fine to implement the program as it currently exists, perhaps using a script to tie the various features of Mantid and an implementation of the algorithms together.
I’ve attached a set of runs that contain a field scan with two period data, together with the program output (both text – ‘output.txt’ – and graphics – ‘pgplot.pdf’).
